20170525 왕진영 AWS 분산딥러닝
- 1. May, 25th, 2017 JinYeong Wang Parallel Computing Lab. Mechanical Engineering Hanyang University AWS Distributed Deep Learning
- 2. 기본 인스턴스 제한 • https://aws.amazon.com/ko/ec2/faqs/#How_many_instances_can_I_run_in_Amazon_EC2
- 3. 기본 인스턴스 제한 숫자 증가 요청 • https://aws.amazon.com/contact-us/ec2-request
- 4. 인스턴스 제한 증가 요청을 하지 않았을 시 볼 수 있는 에러
- 5. Create New Stack in CloudFormation
- 6. Select Template • Template from https://github.com/awslabs/deeplearning-cfn
- 7. Specify Details • CIDR IP from http://www.ipaddressguide.com/cidr#range
- 8. Skip Options
- 9. Review
- 10. Create Complete
- 11. EC2 – Auto Scaling Groups
- 12. Master Instance
- 13. Master Instance IP and DNS
- 19. Run MXNet Distributed Learning • #terminate all running Python processes across workers while read -u 10 host; do ssh $host "pkill -f python" ; done 10<$DEEPLEARNING_WORKERS_PATH • #navigate to the mnist image-classification example directory cd ~/src/mxnet/example/image-classification • #run the MNIST distributed training example ../../tools/launch.py -n $DEEPLEARNING_WORKERS_COUNT -H $DEEPLEARNING_WORKERS_PATH python train_mnist.py --gpus $(seq -s , 0 1 $(($DEEPLEARNING_WORKER_GPU_COUNT - 1))) --network lenet --kv-store dist_sync
- 20. Reference • AWS를 통한 분산 딥러닝 구성하기 – https://aws.amazon.com/ko/blogs/korea/distributed-deep-learning-made-easy/ • Github AWS Distributed Deep Learning Templete – https://github.com/awslabs/deeplearning-cfn • Securely Connect to Linux Instances Running in a Private Amazon VPC – https://aws.amazon.com/ko/blogs/security/securely-connect-to-linux-instances-running-in-a-private- amazon-vpc/ • Putty를 사용하여 Windows에서 Linux 인스턴스에 연결 – http://docs.aws.amazon.com/ko_kr/AWSEC2/latest/UserGuide/putty.html • CIDR IP Address Guide – http://www.ipaddressguide.com/cidr#range